Content Rules

1. What is prohibited

You may not create, attempt to create or upload as a reference any material on the following topics — in a prompt, image, video or audio:

• Sexual content and nudity. Pornography, explicit scenes, sexual acts, nude bodies, erotica, "undressing" people in photos.
• Any content involving minors in a sexual or dangerous context. An absolute ban with no exceptions, including drawn, "anime" and "fantasy" characters.
• Violence and gore. Realistic killings, torture, mutilation, dismemberment, blood and graphic detail, animal cruelty.
• Terrorism and extremism. Terrorist attacks, mass shootings, symbols and propaganda of banned organizations, calls to violence.
• Weapons, explosives, drugs. Manufacturing instructions and diagrams, advertising and sale of illegal substances.
• Self-harm and suicide. Depiction, encouragement or instructions.
• Hate and discrimination. Demeaning people based on nationality, race, religion, gender, age or disability.
• Real people without their consent. Deepfakes, intimate or compromising images, faked voices and appearances, politicians and public figures in fabricated situations. Humiliating, offensive or mocking edits of photos of acquaintances and other private individuals, including pictures and screenshots from social networks and messengers.
• Fraud and forgery. Documents, money, receipts, screenshots of chats and news passed off as real; scams.
• Third-party rights. Exact copies of protected characters, logos and brands — some models reject them.
• Anything that violates the laws of Ukraine or of the user’s country of residence.

Workaround wording, deliberate typos, translation into another language and an “artistic” context do not change a prohibited topic.

2. How violations are counted

A violation is any generation that:

• was rejected by our automatic request filter — such a request is never sent to the model;
• was rejected by the model’s safety filter because of an attached file — an image, video or audio;
• was hidden by the model, which returned an empty (black) result.

If a model’s filter rejected only the text of the prompt, no violation is counted automatically: models’ filters vary in strictness and sometimes trigger on harmless descriptions. Try different wording or another model. The administration sees such rejections and may take them into account in case of obvious attempts to bypass the rules.

Credits for rejected generations are returned to your balance. After each violation, the service tells you how many attempts are left.

3. Blocking

After 5 violations within 30 days, the account is blocked automatically: generations and payments become unavailable, and the remaining credits are not refunded. The administration may block an account without warning for attempts to create material involving minors or clearly illegal content.

Generations are also paused automatically if models reject 8 or more of your requests within 60 minutes: this is what trying wordings to bypass filters looks like. The administration reviews such blocks.

Model filters sometimes make mistakes — for example, on ordinary portraits. If you believe your account was blocked by mistake, email [email protected] — we review every case manually.